Amino Acids (2008) 35: 233–241 DOI 10.1007/s00726-007-0581-5 Printed in The Netherlands

Correlation of in vitro and in vivo models for the oral absorption of peptide drugs F. Fo¨ger, A. Kopf, B. Loretz, K. Albrecht, and A. Bernkop-Schnu¨rch Department of Pharmaceutical Technology, Institute of Pharmacy, Leopold-Franzens University Innsbruck, Innsbruck, Austria Received September 14, 2006 Accepted December 12, 2006 Published online August 30, 2007; # Springer-Verlag 2007

Summary. The aim of this study was to evaluate two in vitro models, Caco-2 monolayer and rat intestinal mucosa, regarding their linear correlation with in vivo bioavailability data of therapeutic peptide drugs after oral administration in rat and human. Furthermore the impact of molecular mass (Mm) of the according peptides on their permeability was evaluated. Transport experiments with commercially available water soluble peptide drugs were conducted using Caco-2 cell monolayer grown on transwell filter membranes and with freshly excised rat intestinal mucosa mounted in Using type chambers. Apparent permeability coefficients (Papp) were calculated and compared with in vivo data derived from the literature. It was shown that, besides a few exceptions, the Mm of peptides linearly correlates with permeability across rat intestinal mucosa (R2 ¼ 0.86; y ¼
196.22x þ 1354.24), with rat oral bioavailability (R2 ¼ 0.64; y ¼
401.90x þ 1268.86) as well as with human oral bioavailability (R2 ¼ 0.91; y ¼
359.43x þ 1103.83). Furthermore it was shown that Papp values of investigated hydrophilic peptides across Caco-2 monolayer displayed lower permeability than across rat intestinal mucosa. A correlation between Papp values across rat intestinal mucosa and in vivo oral bioavailability in human (R2 ¼ 0.98; y ¼ 2.11x þ 0.34) attests the rat in vitro model to be a very useful prediction model for human oral bioavailability of hydrophilic peptide drugs. Presented correlations encourage the use of the rat in vitro model for the prediction of human oral bioavailabilities of hydrophilic peptide drugs. Keywords: Peptides – Oral drug delivery – In vitro in vivo correlation – Caco-2 – Rat intestinal mucosa

1. Introduction Due to recent advances in recombinant biotechnology, many therapeutic peptides and proteins have become readily available (Sayani and Chien, 1996). But non-invasive delivery of these important therapeutic compounds is still limited due to their low bioavailability. Therefore most therapeutic proteins have to be delivered via invasive routes of administration leading to pain and discom-

fort for the patients. Due to the progress in large-scale manufacturing of peptide drugs, convenient application routes gained more and more interest. Among them, oral administration is one of the most accepted ways of dosing for the patients. However, administration of most peptides via the oral route is limited because of their low permeability across intestinal mucosa (BernkopSchnu¨rch and Go¨ckel, 1997). Besides degradation in the gastrointestinal tract, factors such as high molecular mass, hydrophilicity, and the tendency to undergo aggregation account for the low oral bioavailability (Shah et al., 2002) of peptide drugs. For the development of efficacious oral drug delivery systems (Walter et al., 1996) the estimation of oral drug absorption in human based on in vitro and in vivo animal studies is of considerable importance. Therefore, Caco-2 cell monolayers, a human intestinal cell line, and freshly excised rat intestinal mucosa mounted in Ussing type chambers are generally accepted as primary in vitro absorption screening models. There have been several trials to predict human in vivo absorption with the use of Caco-2 and rat intestinal permeation models (LevetTrafit et al., 1996; Walter et al., 1996; Yee, 1997; Lau et al., 2004). Besides primary in vitro studies, animal in vivo pharmacokinetics play a major role in predicting oral in vivo bioavailability in human (Lau et al., 2004), during the process of drug development. However animal studies alone may not always be sufficient for predicting drug absorption in human, especially in the case of degradable peptides, by reason of different intestinal enzymes in animal and human. Because of this, a comparison of different absorption models like the in vitro

234

F. Fo¨ger et al.

model Caco-2, derived from human colorectal carcinoma, with animal in vitro and in vivo studies might be useful. The selection of 21 all perfectly water soluble peptides with therapeutic potential were based on the availability of in vivo data of oral bioavailability in rat or human. The purpose of the present study was to evaluate correlations between Caco-2 cell monolayer, freshly excised rat intestinal mucosa mounted in Ussing type chambers and in vivo bioavailabilities of peptides in rat and human. Additionally we attempted to correlate the molecular mass of the evaluated therapeutic proteins with Papp and in vivo data.

2. Materials and methods 2.1 Data set The in vivo oral bioavailabilities in rat or human as well as in vitro Papp coefficients across Caco-2 monolayer and rat intestinal mucosa mounted in Ussing-type chambers for 21 water soluble peptide drugs (Table 1) were taken from 40 references. The used peptides cover a relatively wide range of molecular mass, ranging from 335 Da to 22 kDa. From Table 1, it can be seen that for some peptide drugs, Caco-2 Papp values from different sources, displaying obvious variations, exist. However, Artursson et al. already showed that results obtained with Caco-2 monolayers vary due to factors such as culture time, passage number and used culture medium (Artursson and Borchardt, 1997). Additionally, in the references indicated TEER values of the Caco-2 monolayers strongly varied, displaying TEER values from 300 to 1200
=cm2. In the case of octreotide, and desmopressin where we have found references for Caco-2 Papp coefficients, displaying a huge variation, we decided to perform the permeation again in our laboratory. Furthermore we performed permeation studies across freshly excised rat intestinal mucosa and=or Caco-2 monolayer if the according peptide drug is commercially available and unless data are found in the literature. 2.2 Materials The peptides (Arg)-vasopressin (AVP), buserelin, D-ala-leu-enkephalin, desmopressin, dynorphin E-2078, IRI-695, leuprolide, melanotan II, octreotide, oxytocin, PheAlaVal, PheAlaValAla and thyreotropin-releasing-hormone (TRH) were obtained from Bachem AG (Bubendorf, Switzerland). Parathyroid-hormone 1–34 (PTH 1–34) was custom synthesised by piCHEM R&D (Austria). Human PTH (1–34) specific ELISA kit was obtained from Immunotopics International (San Clemente, CA). Phosphate buffered saline (PBS) contained 8 g of NaCl, 0.2 g of KCl, 1.536 g of Na2HPO4, and 0.2 g of KH2PO4 per liter (pH 6.8). Modified Eagle Medium (MEM) contained 9.66 g=L MEM powder (Sigma), 2.2 g=L sodium-bicarbonate, 2 mmol=L glutamine (Sigma), 100 U=L penicillin (Sigma), 100 U=L streptomycin (Sigma) and 20% of fetal calf serum (FCS) (pH 6.8). N-(2-hydroxyethyl)piperazine-N0 -(2-ethanesulfonic acid) (HEPES) and all other compounds and reagents were obtained from Sigma (Austria). All chemicals were of analytical grade. 2.3 In vitro permeation studies 2.3.1 Permeation studies on freshly excised rat intestinal mucosa For permeability studies non fasting male Sprague Dawley rats weighting between 250 and 300 g were used. After sacrificing the rats, the first 20 cm

of the proximal jejunum were immediately removed. The excised intestine was cut into strips of 1.5 cm, rinsed free of luminal contents and mounted in Ussing type chambers (0.64 cm2 surface area) without stripping off the underlying muscle layer. The preheated transport medium, containing 250 mM NaCl, 2.6 mM MgSO4, 10 mM KCl, 40 mM glucose and 50 mM NaHCO3 buffered with 50 mM HEPES pH 6.8 was added to the apical and basolateral side. In order to ensure oxygenation and agitation, a mixture of 95% O2 and 5% CO2 was bubbled through each compartment. The Ussing chambers were then placed in a water bath at 37  C. After a 20 min equilibration period, test compounds dissolved in transport medium in a final concentration of 0.05% (w=v) were added to the donor chamber. After 1, 2 and 3 h, 100 mL samples were taken out from the acceptor chamber. The amount of permeated peptides was analyzed by high performance liquid chromatography (HPLC) (Merck HITACHI) equipped with an L-2200 autosampler. Separation of AVP, D-ala-leu-enkephalin, desmopressin, dynorphin E-2078, IRI-695, leuprolide, melanotan II, oxytocin, PheAlaVal, PheAlaValAla and TRH was carried out by using a 25 cm reversed phase Nucleosil 5C18 column (Seibersdorf GmbH). The mobile phase used for separation of these peptides consisted of A: 0.1% TFA in water and B: 0.1% TFA in acetonitrile (90% A to 20% A in 18 min). Separation of octreotide was carried out by using a 15 cm Supelcosil+ LC8-DB column (Sigma-Aldrich) with a mobile phase consisted of 77% A (0.5% (m=v) tetramethylammoniumhydroxid pH 2.7) and 23% B (acetonitrile). A flow rate of 1 mL=min was performed for all analyzed peptides. The peptides AVP, D-ala-leu-enkephalin, desmopressin, dynorphin E-2078, IRI-695, leuprolide, melanotan II, octreotide, oxytocin, PheAlaVal, PheAlaValAla and TRH were monitored by using a diode array detector (L-2450 Merck HITACHI) at 210, 254, 210, 210, 210, 220, 210, 225, 214, 210, 210 and 215 nm, respectively. PTH 1-34 was analysed by using a specific ELISA kit. Calculations were done by interpolations from standard curves obtained with increasing amounts of the according peptide drugs. 2.3.2 Permeation studies on Caco-2 cell monolayer Caco-2 cells (passage number 70) were seeded onto 12 well Transwell polyester membranes (Transwell+, COSTAR, 0.4 mm pore size, 12 mm diameter). The cells were cultured in MEM supplemented with 20% fetal calf serum (FCS). The culture medium was exchanged every other day and the cells were stored in a 5% CO2-incubator at 37  C. Permeation studies were performed with monolayers cultured for 24 days. Caco-2 cell monolayers with trans-epithelial electrical resistance (TEER) values in the range of 400–500
=cm2 were used for the permeation studies. The transport medium contained 250 mM NaCl, 2.6 mM MgSO4, 10 mM KCl, 40 mM glucose and 50 mM NaHCO3 buffered with 50 mM HEPES pH 6.8. Prior to all experiments, each monolayer was washed with phosphate buffer saline (PBS). Then 1 mL of transport medium was added to the apical and 1.5 mL to the basolateral compartment. After a 20 min equilibration period in the 5% CO2 incubator TEER was measured again to assure integrity of the monolayers. An appropriate volume of buffer from the donor chamber was replaced by a solution of the according peptide in a final concentration of 0.05% (m=v). During the experiment the Transwell plates were stored in an incubator. After 1, 2 and 3 h, 100 mL samples were taken out from the acceptor chambers and replaced by preheated buffer. The amount of permeated peptides was analyzed by HPLC as already described above. 2.3.3 Determination of the TEER The integrity of the monolayer was evaluated by measuring the transepithelial electrical resistance (TEER) using an EVOM+ (World Precision Instruments, Sarasota, USA) connected with a pair of electrodes. Caco-2 cell monolayers with trans epithelial electrical resistance (TEER) values in the range of 400–500
cm2 were used for the permeation studies. The TEER was measured before and after the transport studies to ensure integrity of the monolayer during the experiment.

Correlation of in vitro and in vivo models for the oral absorption of peptide drugs

235

Table 1. Summary of in vitro permeability coefficients across rat intestinal mucosa or Caco-2 monolayer and in vivo oral bioavailability in rat or human. Displayed data were observed from the literature as well as from transport experiments performed in our laboratory (indicated by x) Compound

Mm [Da]

Papp rat intestinal mucosa (10
6 )

Papp Caco-2 monolayer (10
6 )

Fabs Rat in vivo [%]

Fabs human in vivo [%]

PheAlaVal TRH

335.4 362.4

5.8  1.8x 4.7  0.9x

2.8 (He, 1996) 1.5 (Yokohama et al., 1984)

– 2.0 (Yokohama et al., 1984)

Azetirelin

384.4 406.5 483

DMP 728

561

1.6 (Sasaki et al., 1997) 1.1 (He et al., 1996) 2.0 (Adusumalli et al., 1996) 2–4 (Burcham et al., 1995)

–

PheAlaValAla IRI-695

4.6  0.46 (Yamamoto, 2001) 4.5  0.4x 4.9  0.4x

2.0  0.5x 2.05  0.44 (Urayama et al., 2003) 1.9  0.3x 2.32  0.76 (Urayama et al., 2003) 1.5  0.5x 0.1  0.3x

D-ala-leu-enkephalin

569.7

Metkephamid

601

Hexarelin

887.0

Oxytocin

1007.2

Octreotide

1019.2

Melanotan II Dynorphin E-2078 Desmopressin

1024 1035 1069.2

2.5  0.3x 1.7  0.7x 0.53  0.29 to 1.13  0.48 (Pantzar et al., 1994)

AVP

1084.3

2.2  0.8x

Buserelin

1239.5

1.0  0.3x

Leuprolide

1269.5

1.6  0.4x

Calcitonin salmon

3431.9

PTH 1–34 Insulin

4117.7 5700

1.91  0.45 (Sinko, 1999) 0.0002x 0.49  0.1 (Yamamoto, 2001)

hGH

22 124

1.25  0.08 to 3.21  0.39 (Aungst and Saitoh, 1996) 3.01  0.39 (Uchiyama et al., 1998) – 1.3  0.9 (Fagerholm et al., 1998) 1.89  0.322 2.09  0.19 (Lundin et al., 1991) 1.5  0.3x

3.45  0.34 (Stoll et al., 2000)

0.35  0.19 (Ribadeneira et al., 1996) 0.7  0.2x 0.049  0.008 (Lang et al., 1997) <0.15 (Westberg et al., 2001) (below detection limit)

0.4 (Lee and Amidon, 2002) 0.22 (Lipka et al., 1996) –

– – –

– – 0.3 (Deghenghi and Camanni, 1994)

1.6  0.3x

–

–

1.7  0.5 (Michael et al., 2000) 1.3  0.7x 0.008  0.004 (Fricker et al., 1996) 2.4  0.3x 1.3  0.2x 0.13 (Artursson and Karlsson, 1991) 4.8  1.6 (Michael et al., 2000) 2.7  0.4x 1.4  0.7x

0.28 (Michael et al., 2000)

0.6 (Drewe et al., 1993)

4.6 (Lan et al., 1994) 0.7 (Murahashi et al., 1989) 0.92 (Michael et al., 2000)

– – 0.1 (Fjellestad et al., 1993)

0.68–0.93 (Miyazaki et al., 2000) 0.8 (Hou et al., 2004), 0.1 (Luessen et al., 1996) 0.17–0.58 (Zheng et al., 1999) 0.022 (Sinko, 1995)

–

0.04  0.01 (Thanou et al., 2000) 0.52  0.02 (Guo et al., 2004) 0.17  0.03 (Song, 2002) 0.002x 0.007 (Ichikawa, 2003)

1.34  0.19 (Wu and Robinson, 1999)

2.3.4 Data analyses The apical to basolateral (AP to BL) permeability coefficients (Papp) of the according peptides were calculated according to the following equation Papp ¼ Q=A  c  t, where Papp is the apparent permeability coefficient (cm=sec), Q is the total amount of peptide permeated after 3 h (mg), A is the diffusion area of the Ussing chamber (0.64 cm2) or of the Caco-2 monolayer (1.13 cm2), c is the initial concentration of according peptide in

– 0.021 (Yamamoto, 2001) 0.25 (Eaimtrakarn et al., 2002) 1.0 (Moore et al., 1986)

– – 0.5–1.4 (Buclin, 2002) – <1 (Shah, 2002)

–

the donor chamber (mg=cm3), and t is the total time of the experiment (sec). The results are reported as means  SD of at least three trials.

2.4 Statistics of the correlation models The statistics of the bivariate correlations were tested by their linear correlation coefficient (R2).

236

F. Fo¨ger et al.

3. Results and discussion 3.1 Impact of molecular mass The high molecular mass of peptide drugs is considered as one of the main obstacles associated with absorption after oral administration. As for hydrophilic peptides a mainly paracellular diffusion through the tight junctions across the gastrointestinal barrier is estimated, and as permeation through these size restricted aqueous pores depends on the molecular mass (Mm) of drugs (Hou et al., 2004), a correlation of Papp and Mm of the according peptides should be expected. Therefore we attempted to correlate the molecular mass of the evaluated peptides with Papp and in vivo data. 3.1.1 Correlation between Mm with rat intestinal Papp When all rat in vitro permeability coefficients of the peptide drugs from Table 1 are plotted as a function of Mm, no linear correlation can be found. However, if peptide drugs with a Mm> 3000 (calcitonin, PTH, insulin and hGH) were excluded, a strong correlation with R2 ¼ 0.86; y ¼
196.22x þ 1354.24 was observed (Fig. 1). These observations indicate that salmon calcitonin (Mm of 3432), insulin (Mm of 5700) and hGH (Mm of 22124) demonstrate higher permeability through rat intestinal mucosa, in relation to their molecular mass, as it might be anticipated. On the one hand, specific peptide transporters could account for such high Papp values, and on the other hand high molecular flexibility of polypeptides are suggested to strongly influence paracellular transport (Salamat-Miller and Johnston, 2005). Paracellular transport for a charged polypeptide even with a Mm of 26.6 kDa, in absence of tight junction modulators, has been reported (Salamat-

Fig. 2. Correlation between permeability coefficients across Caco-2 monolayer and molecular mass of the according peptide drugs. No linear correlation indicated by R2 was observed

Miller and Johnston, 2005). This report challenges the overall assumption that high molecular mass compounds are not able to passively cross intestinal membranes via the tight junctions. So far no specific peptide transporter in the gastrointestinal tract for insulin, calcitonin or hGH has been reported. However, it was reported that besides paracellular transport other routes may be involved in insulin absorption (Lane and Corrigan, 2006). Derossi et al. reported transcellular receptor-independent absorption of several peptides (Derossi et al., 1998). 3.1.2 Correlation of Mm and Caco-2 Papp No linear correlation was observed by plotting all 21 peptide drugs. Even when peptides with Mm > 3000 were excluded, no linear correlation could be found (Fig. 2). Also if just the Caco-2 Papp values from our laboratory (n ¼ 12) were correlated versus Mm, no linear correlation was found. Several studies have shown good correlations by using this human colorectal carcinoma cell line, however, large interlaboratory differences in the cultivation of Caco-2 monolayer and other reasons such as poor paracellular permeability across Caco-2 monolayer might explain why no linear correlations were found between Mm and Caco-2 Papp. 3.1.3 Correlation of Mm and rat oral bioavailability

Fig. 1. Correlation between permeability coefficients across rat intestinal mucosa mounted in Ussing-type chambers and molecular mass of the according peptide drugs. The best fit line is based on a linear regression analysis indicated with R2

When rat in vivo bioavailability of all peptide drugs with Mm < 3000 from Table 1 were plotted as a function of Mm, no correlation was observed. However, when the Pglycoprotein (P-gp) substrates D-ala-leu-enkephalin, metkephamid (white squares in Fig. 3) and the cyclic peptide melanotan II were excluded, a moderate to strong correlation with R2 ¼ 0.64; y ¼
401.90x þ 1268.86 was found. P-gp, an apically polarized efflux pump, which is located

Correlation of in vitro and in vivo models for the oral absorption of peptide drugs

237

ered. So, based on increasing human in vivo data, this prediction model will need to be continuously refined. 3.2 Prediction models based on Caco-2 monolayer

Fig. 3. Correlation between in vivo oral bioavailability in the rat and molecular mass of the according peptide drugs. The best fit line is based on a linear regression analysis indicated with R2. The P-gp substrates metkephamid and D-ala-leu-enkephalin (white square) were excluded from the plot demonstrating a significantly lower oral bioavailability

in the membrane of enterocytes (Troutman and Thakker, 2003) limits the oral bioavailability of a lot of structurally diverse compounds such as enkephalin-analogues and several other substances (Kim, 2002) by translocating substrates from the inner side of the membrane to the outer side. In Fig. 3 it can be seen that the P-gp substrates D-ala-leu-enkephalin and metkephamid demonstrate a significantly lower oral bioavailability in the rat as compared with non-P-gp substrates. In contrast, the cyclic heptapeptide melanotan II displayed significantly higher oral bioavailability (Fabs ¼ 4.6%). The high apparent partition coefficient of 2.82 (n-octanol=water at pH 7.35) might explain the high oral bioavailability of melanotan II (Lan et al., 1994). 3.1.4 Correlation of Mm and human oral bioavailability A strong linear correlation with R2 ¼ 0.91; y ¼
359.43 þ 1103.83 between the Mm of peptide drugs and oral bioavailability in human is demonstrated in Fig. 4. However, the poor available human in vivo data should be consid-

Fig. 4. Correlation between in vivo oral bioavailability in human and molecular mass of the according peptide drugs. The best fit line is based on a linear regression analysis indicated with R2

Transport studies across Caco-2 monolayer represent one of the most accepted in vitro permeation models. Several studies showed good correlations between Caco-2 permeability and the fraction of drug absorbed after oral administration (Lau et al., 2004). However limitations such as the lack of intestinal phase 1 metabolic enzymes like CYP3A4 and furthermore the absence of villi and mucuslayer have to be noted. Additionally, Caco-2 cells represent tighter junctions in comparison with human or animal small intestine, which is explained by the colonic origin of these cells (Matsson et al., 2005). Due to huge interlaboratory discrepancies, a direct comparison of permeability of the same drugs might often fail. As shown in Table 1, big variations for the peptide drugs desmopressin and octreotide have been found. However, after repeated permeation studies across Caco-2 monolayer performed in our laboratory, it was possible to exclude an obvious outlier of octreotide (given with a permeability coefficient of 0.008  10
6 ). In the case of octreotide we used the Papp value evaluated in our laboratory for the correlations. 3.2.1 Correlation of Caco-2 and rat intestinal Papp Just a poor linear correlation with R2 ¼ 0.48; y ¼ 0.35x þ 0.19 between Caco-2 and rat intestinal mucosa Papp values can be seen in Fig. 5. However, this correlation clearly shows that hydrophilic peptides exhibit significantly higher permeability across rat intestinal mucosa than across the cultured Caco-2 monolayers. As reported earlier, permeability of hydrophilic peptide drugs across Caco-2 monolayer is lower compared to permeability

Fig. 5. Correlation between permeability coefficients across rat intestinal mucosa mounted in Ussing-type chambers and Caco-2 monolayer. The best fit line is based on a linear regression analysis indicated with R2

238

F. Fo¨ger et al.

Fig. 6. Correlation between permeability coefficients across Caco-2 monolayer and in vivo oral bioavailability in the rat. The best fit line is based on a linear regression analysis indicated with R2

through intestinal mucosa (Walter et al., 1996). This was explained on the one hand by a reduced absorption area due to the lack of villi and on the other hand by lower paracellular permeability through the tighter junctions in Caco-2 monolayer. Besides the Caco-2 model, MadinDarby canine kidney (MDCK) cells have been reported as model for the permeation of passively absorbed drugs (Tang et al., 2002), however, in regard to oral drug delivery, Caco-2 cells are used more frequently. 3.2.2 Correlation of Caco-2 Papp and rat oral bioavailability A correlation with R2 ¼ 0.48; y ¼ 0.48x þ 0.58 was observed after plotting Caco-2 Papp values versus rat in vivo bioavailability (Fig. 6). As mentioned above, different sources of Caco-2 Papp values might account for this moderate correlation. Furthermore, it has already been reported that permeability of several drugs across Caco-2 monolayer is lower than in comparison with intestinal mucosa (Walter et al., 1996). Due to their extraction from human colon carcinomas, Caco-2 permeation models are more related to colon permeability.

accepted as primary in vitro absorption screening models. Several drug transporters such as P-gp and MRP efflux pumps are expressed in both, in rat and in human intestinal mucosa. Furthermore Mouly et al. reported that the quantitative expression of the efflux pump P-gp in the different human intestinal segments is comparable to the rat intestinal parts (Mouly and Paine, 2003). In addition, mRNAs for peptide transporter 1 (PepT1), peptide transporter 3 (PTR3), peptide=histidine transporter 1 (PHT1) and the human peptide transporter 1 (HPT-1) were widely expressed in the rat gastro-intestinal-tract (Herrera-Ruiz et al., 2001). Therefore the rat model seems to be suitable for several compounds in predicting drug absorption in human. 3.3.1 Correlation of rat intestinal Papp and rat oral bioavailability A plot of rat intestinal Papp versus rat in vivo oral bioavailability showed a strong correlation with R2 ¼ 0.81; y ¼ 1.89x þ 1.27 (Fig. 7). The outliers desmopressin, DMP and melanotan II were excluded from this correlation because they obviously showed higher in vivo bioavailabilities as it should be estimated from their in vitro data. The cyclic peptide DMP was reported to be a substrate of P-gp (Saitoh and Aungst, 1997) however, P-gp should be expressed in both, in the rat in vitro as well as in the rat in vivo model. One explanation for the high oral in vivo bioavailability of DMP and melanotan II could be their cyclization, because conformational changes can influence membrane permeability (Saitoh and Aungst, 1997). Otherwise the reported high stability of DMP in

3.2.3 Correlation of Caco-2 Papp and human oral bioavailability Due to the poor available data, no attempts to correlate Caco-2 Papp with human oral bioavailability of hydrophilic peptide drugs were made. Furthermore, there have been reports that Caco-2 monolayer can only be used to predict passive transcellular transport and not paracellular transport, by reason of smaller pore size resembling to the colonic epithelia (Yee, 1997). 3.3 Prediction models based on rat intestinal Papp Permeation studies across freshly excised rat intestinal mucosa mounted in Ussing type chambers are generally

Fig. 7. Correlation between permeability coefficients across rat intestinal mucosa mounted in Ussing-type chambers and in vivo oral bioavailability in the rat. The best fit line is based on a linear regression analysis indicated with R2. The outliers desmopressin, the cyclic P-gp substrate DMP (white squares) and the cyclic heptapeptide melanotan II were excluded from this linear correlation because they represent noticeable higher in vivo bioavailability in relation to their in vitro Papp values

Correlation of in vitro and in vivo models for the oral absorption of peptide drugs

239

gastric and intestinal fluid as well as in plasma (Aungst and Saitoh, 1996), and the high apparent partition coefficient of melanotan II could explain their proportional high oral bioavailability. In the case of the second outlier desmopressin, the existence of an active transport mechanism was reported to be unlikely, however Pantzar et al. confirmed the existence of regional absorption differences in vitro as well as in vivo in rats (Pantzar et al., 1995).

dictors then rats, guinea pigs or dogs for human small intestinal metabolism of CYP3A4 substrates. However, the usefulness of the rat model for predictions of oral bioavailability of peptide drugs in human was demonstrated in Fig. 7. A direct linear correlation of rat and human in vivo data, however, is yet not meaningful, due to limited available data.

3.3.2 Correlation of rat intestinal Papp and human oral bioavailability

4. Conclusion

2

A very strong linear correlation with R ¼ 0.98; y ¼ 2.11x þ 0.34 between the rat in vitro permeation model and human oral in vivo bioavailabilities was found (Fig. 8). However it should be noted that the data set is limited because of poor available human in vivo data. Nevertheless such a strong linear relationship encourages the use of the rat model for the prediction of human oral bioavailabilities of hydrophilic peptide drugs.

Although, recent studies challenged the impact of Mm on paracellular transport, our study demonstrates that besides a few exceptions, the Mm of peptides linearly correlates with their permeability. For Papp values of hydrophilic peptides across Caco-2 monolayer higher permeability of the same compound across rat intestinal mucosa or in vivo in the rat might be expected. Finally, the presented correlations attest the rat in vitro model to be a better prediction model for human oral bioavailabilities of hydrophilic peptide drugs in comparison with the Caco-2 monolayer.

3.4 Correlation of rat and human oral bioavailabilities Several animal models are used to predict intestinal drug absorption in human. Frequently used models are rats, mice, guinea pigs, pigs and dogs. In a comparison concerning the paracellular transport of hydrophilic macromolecules, rats appeared to be better predictors than dogs for human drug absorption (He et al., 1998). In an investigation of different species of animals (monkeys, guinea pigs, dogs and rats) concerning their intestinal enzymatic activity, enterokinase activity in the rat duodenal mucosa was found to resemble closest the activity in the human duodenum (Malis et al., 1977). Contrarywise, Komura et al. (2002) implied that monkeys would be better pre-

Fig. 8. Correlation between permeability coefficients across rat intestinal mucosa mounted in Ussing-type chambers and in vivo oral bioavailability in human. The best fit line is based on a linear regression analysis indicated with R2

References Adusumalli V, Corkum N, Jacala A, Mukherjee T, Goodlett D, Crowther J, McConnell I, Goldstein G (1996) Pharmacokinetics and toxicokinetics of an orally active tripeptide, IRI-695, in animals. Biopharm Drug Disposition 17: 25–41 Artursson P, Borchardt RT (1997) Intestinal drug absorption and metabolism in cell cultures: Caco-2 and beyond. Pharm Res 14: 1655–1658 Artursson P, Karlsson J (1991) Correlation between oral drug absorption in humans and apparent drug permeability coefficients in human intestinal epithelial (Caco-2) cells. Biochem Biophys Res Commun 175: 880–885 Aungst BJ, Saitoh H (1996) Intestinal absorption barriers and transport mechanisms, including secretory transport, for a cyclic peptide, fibrinogen antagonist. Pharm Res 13: 114–119 Bernkop-Schnu¨rch A, Go¨ckel NC (1997) Development and analysis of a polymer protecting from luminal enzymatic degradation caused by aChymotrypsin. Drug Develop Ind Pharm 23: 733–740 Buclin T, Rochat MC, Burckhardt P, Azira M, Attinger M (2002) Bioavailability and biological efficacy of a new oral formulation of salmon calcitonin in healthy volunteers. J Bone Miner Res 17: 1478–1485 Burcham DL, Aungst BA, Hussain M, Gorko MA, Quon CY, Huang SM (1995) The effect of absorption enhancers on the oral absorption of the GP IIB=IIIA receptor antagonist, DMP728, in rats and dogs. Pharm Res 12: 2065–2070 Deghenghi R, Camanni F (1994) Growth hormone-releasing activity of hexarelin, a new synthetic hexapeptide, after intravenous, subcutaneous, intranasal, and oral administration in man. J Clin Endocrinol Metab 78: 693–698 Derossi D, Chassaing G, Prochiantz A (1998) Trojan peptides: the penetrating system for intracellular delivery. Trends Cell Biol 8: 84–87 Drewe J, Fricker G, Vonderscher J, Beglinger C (1993) Enteral absorption of octreotide: absorption enhancement by polyoxyethylene-24-cholesterol ether. Br J Pharmacol 108: 298–303

240

F. Fo¨ger et al.

Eaimtrakarn S, Rama Prasad YV, Ohno T, Konishi T, Yoshikawa Y, Shibata N, Takada K (2002) Absorption enhancing effect of labrasol on the intestinal absorption of insulin in rats. J Drug Target 10: 255–260 Fagerholm U, Sjostrom B, Sroka-Markovic J, Wijk A, Svensson M, Lennernas H (1998) The effect of a drug-delivery system consisting of soybean phosphatidyl choline and medium-chain monoacylglycerol on the intestinal permeability of hexarelin in the rat. J Pharm Pharmacol 50: 467–473 Fjellestad-Paulsen A, Hoglund P, Lundin S, Paulsen O (1993) Pharmacokinetics of 1-deamino-8-D-arginine vasopressin after various routes of administration in healthy volunteers. Clin Endocrinol (Oxf) 38: 177–182 Fricker G, Fahr A, Beglinger C, Kissel T, Reiter G, Drewe J (1996) Permeation enhancement of octreotide by specific bile salts in rats and human subjects: in vitro, in vivo correlations. Br J Pharmacol 117: 217–223 Guo J, Ping Q, Jiang G, Dong J, Qi S, Feng L, Li Z, Li C (2004) Transport of leuprolide across rat intestine, rabbit intestine and Caco-2 cell monolayer. Int J Pharm 278: 415–422 He YL, Murby S, Gifford L, Collett A, Warhurst G, Douglas KT, Rowland M, Ayrton J (1996) Oral absorption of D-oligopeptides in rats via the paracellular route. Pharm Res 13: 1673–1678 He YL, Murby S, Warhurst G, Gifford L, Walker D, Ayrton J, Easmond R, Rowland M (1998) Species differences in size discrimination in the paracellular pathway reflected by oral bioavailability of poly(ethylene glycol) and D-peptides. J Pharm Sci 87: 626–633 Herrera-Ruiz D, Wang Q, Cook TJ, Knipp GT (2001) Spatial expression patterns of peptide transporters in the human and rat gastrointestinal tracts, Caco-2 in vitro cell culture model, and multiple human tissues. AAPS Pharmsci 3: 1–12 Hou TJ, Zhang W, Xia K, Qiao XB, Xu XJ (2004) ADME evaluation in drug discovery. 5. Correlation of Caco-2 permeation with simple molecular properties. J Chem Inf Comput Sci 44: 1585–1600 Ichikawa H, Peppas NA (2003) Novel complexation hydrogels for oral peptide delivery: in vitro evaluation of their cytocompatibility and insulin-transport enhancing effects using Caco-2 cell monolayers. J Biomed Mater Res A 67: 609–617 Kim RB (2002) Drugs as P-glycoprotein substrates, inhibitors, and inducers. Drug Metab Rev 34: 47–54 Komura H, Yasuda H, Yoshida NH, Sugiyama Y (2002) Species difference in nisoldipine oxidation activity in the small intestine. Drug Metab Pharmacokin 17: 427–436 Lan EL, Ugwu SO, Blanchard J, Fang X, Hruby VJ, Sharma S (1994) Preformulation studies with melanotan-II: a potential skin cancer chemopreventive peptide. J Pharm Sci 83: 1081–1083 Lane ME, Corrigan OI (2006) Paracellular and transcellular pathways facilitate insulin permeability in rat gut. J Pharm Pharmacol 58: 271–275 Lang VB, Langguth P, Ottiger C, Wunderli-Allenspach H, Rognan D, Rothen-Rutishauser B, Perriard JC, Lang S, Biber J, Merkle HP (1997) Structure-permeation relations of met-enkephalin peptide analogues on absorption and secretion mechanisms in Caco-2 monolayers. J Pharm Sci 86: 846–853 Lau YY, Chen YH, Liu T, Li C, Cui X, White RE, Cheng KC (2004) Evaluation of a novel in vitro Caco-2 hepatocyte hybrid system for predicting in vivo oral bioavailability. Drug Metab Dispos 32: 937–942 Lee HJ, Amidon GL (2002) The effect of enzyme inhibitor and absorption site following [D-ala2, D-leu5]enkephalin oral administration in rats. Biopharm Drug Dispos 23: 131–141 Levet-Trafit B, Gruyer MS, Marjanovic M, Chou RC (1996) Estimation of oral drug absorption in man based on intestine permeability in rats. Pharmacol Lett 58: 359–363 Lipka E, Crison J, Amidon GL (1996) Transmembrane transport of peptide type compounds: prospects for oral delivery. J Control Release 39: 121–129

Luessen HL, de Leeuw BJ, Langemeyer MW, de Boer AB, Verhoef JC, Junginger HE (1996) Mucoadhesive polymers in peroral peptide drug delivery. VI. Carbomer and chitosan improve the intestinal absorption of the peptide drug buserelin in vivo. Pharm Res 13: 1668–1672 Lundin S, Pantzar N, Broeders A, Ohlin M, Westrom BR (1991) Differences in transport rate of oxytocin and vasopressin analogues across proximal and distal isolated segments of the small intestine of the rat. Pharm Res 8: 1274–1280 Malis F, Lojda Z, Stepankova R, Slezak Z (1977) Intestinal gradient of enterokinase activity in different species of animals. Acta Univ Carol Med Monogr 77: 113–117 Matsson P, Bergstro¨m CAS, Nagahara N, Tavelin S, Norinder U, Artursson P (2005) Exploring the role of different drug transport routes in permeability screening. J Med Chem 48: 604–613 Michael S, Thole M, Dillmann R, Fahr A, Drewew J, Fricker G (2000) Improvement of intestinal peptide absorption by a synthetic bile acid derivatice, cholylsarcosine. Eur J Pharm Sci 10: 133–140 Miyazaki M, Swada S, Nishide T, Iwanga K, Morimoto K, Kakemi M (2000) Bioavailability assessment of arginine-vasopressin (AVP) using pharmacokinetic-pharmacodynamic (PK-PD) modelling in the rat. Biol Pharm Bull 23: 87–96 Moore JA, Pletcher SA, Ross M (1986) Absorption enhancement of growth hormone from the gastrointestinal tract of rats. Int J Pharm 34: 35–43 Mouly S, Paine MF (2003) P-glycoprotein increases from proximal to distal regions of human small intestine. Pharm Res 20: 1595–1599 Murahashi N, Kato A, Koyama N, Watanabe S, Yuzuriha T, Miyake Y (1989) Rectal absorption of E-2078 (dynorphin analogue peptide) in rats. J Pharm Pharmacol 41: 770–774 Pantzar N, Lundin S, Wester L, Westrom BR (1994) Bidirectional smallintestinal permeability in the rat to some common marker molecules in vitro. Scand J Gastroenterol 29: 703–709 Pantzar N, Lundin S, Westrom BR (1995) Different properties of the paracellular pathway account for the regional small intestinal permeability to the peptide desmopressin. J Pharm Sci 84: 1245–1248 Ribadeneira MD, Aungst BJ, Eyermann CJ, Huang SM (1996) Effects of structural modifications on the intestinal permeability of angiotensin II receptor antagonists and the correlation of in vivo, in situ, and in vivo absorption. Pharm Res 13: 227–233 Roumi M, Kwong E, Deghenghi R, Locatelli V, Marleau S, Du Souich P, Beliveau R, Ong H (2001) Permeability of the peptidic GH secretagogues hexarelin and EP 51389, across rat jejunum. Peptides 22: 1129–1238 Saitoh H, Aungst BJ (1997) Prodrug and analog approaches to improving the intestinal absorption of a cyclic peptide, GPIIb=IIIa receptor antagonist. Pharm Res 14: 1026–1029 Salamat-Miller N, Johnston TP (2005) Current strategies used to enhance the paracellular transport of therapeutic polypeptides across the intestinal epithelium. Int J Pharm 294: 201–216 Sasaki I, Tamura T, Shibakawa T, Fujita T, Murakami M, Yamamoto A, Muranishi S (1997) Metabolism of azetirelin, a new thyrotropinreleasing hormone (TRH) analogue, by intestinal microorganisms. Pharm Res 14: 1004–1007 Sayani AP, Chien YW (1996) Systemic delivery of peptides and proteins across absorptive mucosa. Crit Rev Ther Drug Carrier Syst 13: 85–184 Shah RB, Ahsan F, Khan MA (2002) Oral delivery of proteins: progress and prognostication. Crit Rev Ther Drug Carrier Sys 19: 135–169 Sinko PJ, Chung CL, Mc Whorter LT, Stern W, Wagner E, Gilligan JP (1995) Utility of pharmacodynamic measures for assessing the oral bioavailability of peptides. 1. Administration of recombinant salmon calcitonin in rats. J Pharm Sci 84: 1374–1378 Sinko PJ, Lee YH, Makhey V, Leesman GD, Sutyak JD, Yu H, Perry B, Smith CL, Hu P, Wagner EJ, Falzone LM, Mc Whorter LT, Gilligan JP, Stern W (1999) Biopharmaceutical approaches for developing and assessing oral peptide delivery strategies and systems: in vitro perme-

Correlation of in vitro and in vivo models for the oral absorption of peptide drugs ability and in vivo oral absorption of salmon calcitonin (sCT). Pharm Res 16: 527–533 Song KH, Chung SJ, Shim CK (2002) Preparation and evaluation of proliposomes containing salmon calcitonin. J Control Release 84: 27–37 Stoll BR, Leipold HR, Milstein S, Edwards DA (2000) A mechanistic analysis of carrier-mediated oral delivery of protein therapeutics. J Control Release 64: 217–228 Tang F, Horie K, Borchardt RT (2002) Are MDCK cells transfected with the human MRP2 gene a good model of the human intestinal mucosa? Pharm Res 19: 773–779 Thanou M, Florea BI, Langemeyer MW, Verhoef JC, Junginger HE (2000) N-trimethylated chitosan chloride (TMC) improves the intestinal permeation of the peptide drug buserelin in vitro (Caco-2 cells) and in vivo (rats). Pharm Res 17: 27–31 Troutman MD, Thakker DR (2003) Rhodamine 123 requires carriermediated influx for its activity as a P-glycoprotein substrate in Caco-2 cells. Pharm Res 20: 1192–1199 Uchiyama T, Kotani A, Kishida T, Tatsumi H, Okamoto A, Fujita T, Muratami M, Muranishi S, Yamamoto A (1998) Effects of various protease inhibitors on the stability and permeability of [D-Ala2, DLeu5]enkephalin in the rat intestine: comparison with leucine enkephalin. J Pharm Sci 87: 448–452 Urayama A, Yamada S, Deguchi Y, Ohmori Y, Rimura R (2003) Studies on the transport of thyrotropin-releasing hormone (TRH) analogues in Caco-2 cell monolayers. J Pharm Pharmacol 55: 603–608 Walter E, Janich S, Roessler BJ, Hilfinger JM, Amidon GL (1996) HT29MTX=Caco-2 cocultures as an in vitro model for the intestinal epithe-

241

lium: in vitro-in vivo correlation with permeability data from rats and humans. J Pharm Sci 85: 1070–1076 Westberg C, Benkestock K, Fatouros A, Svensson M, Sjostrom B (2001) Hexarelin-evaluation of factors influencing oral bioavailability and ways to improve absorption. J Pharm Pharmacol 53: 1257–1264 Wu SJ, Robinson JR (1999) Transcellular and lipophilic complex-enhanced intestinal absorption of human growth hormone. Pharm Res 16: 1266–1272 Yamamoto A (2001) Improvement of transmucosal absorption of biologically active peptide drugs. Yakugaku Zasshi 121: 929–948 Yee S (1997) In vitro permeability across Caco-2 cells (colonic) can predict in vivo (small intestinal) absorption in man-fact or myth. Pharm Res 14: 763–766 Yokohama S, Yamashita K, Toguchi H, Takeuchi J, Kitamori N (1984) Absorption of thyrotropin-releasing hormone after oral administration of TRH tartrate monohydrate in the rat, dog and human. J Pharmacobiodyn 7: 101–111 Zheng Y, Qiu Y, Lu MF, Hoffman D, Reiland TL (1999) Permeability and absorption of leuprolide from various intestinal regions in rabbits and rats. Int J Pharm 185: 83–92

Authors’ address: Andreas Bernkop-Schnu¨rch, Department of Pharmaceutical Technology, Institute of Pharmacy, Leopold-FranzensUniversity Innsbruck, Innrain 52, Josef Mo¨ller Haus, A-6020 Innsbruck, Austria, Fax: þ43-512-507-2933, E-mail: [email protected]